These changes fixed the horizontal scrolling issues for me. Cursory testing revealed no unintended sideeffects, but more testing should be done. I did not find an easy way to test if horizontal scrolling works as intended, plainly logging the events looks good, however. No image-side changes were necessary, so I would claim that the VMs for the other platforms can continue to work as normal, even with this change just on x11. Using the workaround to mark a scroll event by detecting all modifiers being pressed down instead, will require an image-side change though. You can view, comment on, or merge this pull request online at:https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/303 Commit Summary
File ChangesPatch Links:
— |
Hi Tom,
So what /is/ happening with genuine mouse wheel events? Should we instead use e.g.
— |
In reply to this post by David T Lewis
I have found that the image had good support for MouseWheelEvents for a while. Based on the method stamps it seems Marcel added the necessary image-side code on 6/10/2016 via the The original comment that you pointed out in the opensmalltalk sources comes from commit 3010e44 You're right though that images from before October '16 on X11 would find themselves without scroll support. Do the images have some sort of capability advertising system or will this need to be a primitive invoked on system startup? I believe this is a rather urgent fix. As Chris already pointed out on the mailing list, it effectively makes scrolling in any function larger than the code panel effectively impossible on X11 in 5.2. Or at least I keep slipping in horizontal scroll events when using my touchpad. If this PR turns out to be a major undertaking, however, we may be better off reverting the commit that enabled horizontal scroll events in the first place for the moment and revisit this afterwards with the cleanest possible solution. What is the scope of backwards compatibility for the VM? Are breaking changes like this allowed in major version increments (e.g. 5.2->5.3 or 5.2->6.0) or to be avoided at all cost? — |
In reply to this post by David T Lewis
The scope of backward-compatibility for the VM is complete except between major releases, so a 5.n VM must be backward-compatible with any and all 5.x, x <= n. So the right way to do this is Tom, if you have energy can you write code for a) ? I will do b) & c), and will do a) if you're indisposed. — |
In reply to this post by David T Lewis
Closed in favor of 897ef17 — |
In reply to this post by David T Lewis
Closed #303. — |
In reply to this post by David T Lewis
I see you were a little faster than me :) Thank you for all the adaptions! — |
Free forum by Nabble | Edit this page |